In this paper, the global stability problem for constrained robot motions in the presence of constraint uncertainties is investigated. We focus on the uncertainties in the constraint functions and their effects on the global stability. PD type controllers are used and conditions for global stability are developed using Lyapunov's direct approach. In the presence of the constraint uncertainties under investigation, the desired position and constraint force can be guaranteed with global asymptotic convergence. The developed conditions for feedback gain selections clearly show the effects of the constraint uncertainties. For the case when the velocity measurements are not available, conditions for global stability regulation are also established and the robot controller uses only the measurements of the position angles. Finally, we consider the case where the robot joints are flexible and global stability conditions are given.

This paper presents a virtual headstick system as an alternative to the conventional passive headstick for persons with limited upper extremity function. The system is composed of a pair of kinematically dissimilar master-slave robots with the master robot being operated by the user's head. At the remote site, the end-effector of the slave robot moves as if it were at the tip of an imaginary headstick attached to the user's head. A unique feature of this system is that through force-reflection, the virtual headstick provides the user with proprioceptive information as in a conventional headstick, but with an augmentation of workspace volume and additional mechanical power. This paper describes the test-bed development, system identification, bilateral control implementation, and system performance evaluation.

This paper presents a design of an assembly sequence planner based on a “plan reuse” philosophy. Most of assembly planning research in the past has attempted to completely plan each problem from scratch. This research shows that stored cases of basic assembly configurations can be applied to a given assembly problem. It is observed that the number of such basic assembly configurations is quite small. The planner divides the assembly into a number of constituent configurations, which are called “loops”. These act as subgoals in its search for solutions. Plans retrieved for all subgoals are fused into a set of plans that are consistent with the constraints implied by each plan. Application specific constraints on the assembly are explicitly handled in the second phase of planning. Mechanisms for assembly representation and implementation details of the planner are also presented.

This paper presents a hybrid approach to optimize the counterweight balancing of a robot arm. A new technique that combines an artificial intelligence technique called the genetic algorithm (GA) and the weighted min-max multiobjective optimization method is proposed. These techniques are included in a system developed by the authors, called MOSES, which is intended to be used as a tool for engineering design optimization. The results presented here show how the new proposed technique can get better trade-off solutions and a more accurate Pareto front for this highly non-convex problem using an ad-hoc floating point representation and traditional genetic operators. Finally, a methodology to compute the ideal vector using a genetic algorithm is presented. It is shown how with a very simple dynamic approach to adjust the parameters of the GA, it is possible to obtain better results than those previously reported in the literature for this problem.

In this paper, we are introducing a system which is able to recognize polyhedral objects in an indoor environment. Our system is intended to be implemented on autonomous mobile platforms in order to enable the localization or research of a precise item. The algorithm is based on the use of geometric quasi-invariants associated to every object. These geometric quasi-invariants correspond to the ratio of the lengths as well as the angle formed by the pair of segments which are in relationship and which are constituting the object. We present some experimental results gained on one of our platforms in our laboratory.

This paper presents a new adaptive controller as a solution to the problem of stabilizing nonholonomic mechanical systems in the presence of incomplete information concerning the system dynamic model. The proposed control system consists of two su bsystems: a slightly modified version of the kinematic stabilization strategy of M'Closkey and Murray which generates a desired velocity trajectory for the nonholonomic system, and an adaptive control scheme which ensures that this velocity trajector y is accurately tracked. This approach is shown to provide arbitrarily accurate stabilization to any desired configuration and can be implemented with no knowledge of the system dynamic model. The efficacy of the proposed stabilization strategy is illustr ated through extensive computer simulations with nonholonomic mechanical systems arising from explicit constraints on the system kinematics and from symmetries of the system dynamics.

The Multimodal User Supervised Interface and Intelligent Control (MUSIIC) project focuses on a multimodal human-machine interface which addresses user need to manipulate familiar objects in an unstructured environment. The control of a robot by individuals with significant physical limitations presents a challenging problem of telemanipulation. This is addressed by a unique user-interface integrating the user's command (speech) and gestures (pointing) with autonomous planning techniques (knowledge-bases and 3-D vision). The resultant test-bed offers the opportunity to study telemanipulation by individuals with physical disabilities, and can be generalized to an effective technique for other, including remote and time-delayed, telemanipulation. This paper focuses on the knowledge-driven planning mechanism that is central to the MUSIIC system.

An efficient algorithm for generating an optimal plan for part-bringing tasks, using robotic manipulators, is introduced. The task of transporting a micro-part in a partially unstructured environment, that includes obstacles whose locations are not initially known, is introduced with the optimal plan formulated on the basis of the observed environmental conditions. Fuzzy set theory, well-suited to the management of uncertainty, is introduced to address the uncertainty associated with the part-bringing procedure. A part-bringing algorithm for generating the optimal plan related to a part assembly, despite existing obstacles, is presented. It is shown that the machine organizer using a sensor system can intelligently determine an optimal plan, based on explicit performance criteria, to overcome environmental uncertainty. The algorithm utilizes knowledge processing functions such as machine reasoning, planning, memory, and decision-making. Simulation results show the effectiveness of the proposed approach. The proposed algorithm is applicable not only to a wide range of robotic tasks including pick and place operations and maneuvering mobile based robots around obstacles, but also to the control of unmanned aircraft.

In this paper a path planning method for off-line programming of a joint robot is described. The method can automatically choose the easiest and safest route for an industrial robot from one position to another. The method is based on the use of a Self Organised Feature Map (SOFM) neural network. By using the SOFM neural network the method can adapt to different working environments of the robot. According to test results one can conclude that the SOFM neural network is a useful tool for the path planning problem of a robot.

The frequent use of past experience by human engineers when solving new problems has led to an interest in the use of case based reasoning (CBR) to help automate engineering design. In engineering design it often occurs that many past exp eriences must be combined to solve a new problem, and thus the process of case based adaptation must efficiently and systematically combine information from many sources. We have developed a constraint based methodology for case combination that allows its application across a wide range of problems. We have shown that our approach provides an efficient adaptation methodology that ensures convergence upon a solution if one exists, provides a uniform representation of cases, and is generalizable beyond just one domain. Our technique is implemented in a case based reasoning system called COMPOSER, which ha s been tested in two design domains: assembly sequence design and configuration design.

The ability of manipulating flexible objects, such as rubber belts and paper sheets, is important in automated manufacturing systems. This paper describes a novel approach to vision-guided assembly of flexible objects. The operation dealt with in this paper is to assemble a rubber belt with fixed pulleys. By analyzing possible states of the belt based on the empirical knowledge of the belt, we can derive a method to have not only the action planning but also the visual verification planning. We have implemented a belt assembly system using the two manipulators and a laser range finder as the sensor, and succeeded in performing the belt-pulley assembly. The extension of our approach to other kinds of assembly of flexible objects is also discussed.

In this paper, nine adaptive control algorithms are compared. The best two of them are tested experimentally. It is shown that the Adaptive FeedForward Controller AFFC) is well suited for learning the parameters of the dynamic equation, even in the presence of friction and noise. The resulting control performance is better than with measured parameters for any trajectory in the workspace. When the task consists of repeating the same trajectory, an adaptive look-up-table MEMory, introduced and analyzed in this paper, is simpler to implement and results in even better control performance.

The minimization of the joint torques based on the ∞-norm is proposed for the dynamic control of a kinematically redundant manipulator. The ∞-norm is preferred to the 2-norm in the minimization of the joint torques since the maximum torques of the actuators are limited. To obtain the minimum ∞-norm torque solution, we devised a new algorithm that uses the acceleration polyhedron representing the end-effector's acceleration capability. Usually the minimization of the joint torques has an instability problem for the long trajectories of the end-effector. To suppress this instability problem, an inequality constraint, named the feasibility constraint, is developed from the geometrical relation between the required end-effector acceleration and the acceleration polyhedron. The minimization of the °-norm of the joint torques subject to the feasibility constraint is shown to improve the performances through the simulations of a 3-link planar redundant manipulator.

A new systematic exploration method is addressed that permits a mobile robot to effectively acquire the information on an unknown environment without wasting time. The algorithm is composed of following three modules: the first is the decomposition of a workspace into several sub-nodes by employing the concept of Quadtree. These nodes are chosen as sub-goals to be reached successively. The second is that controls the robot to follow the boundaries of objects in the environment in order to reach each sub-goal avoiding the situation of the robot entering a local minima. Here we utilize the imaginary distance forces exerted by objects. The third is the node conditions for estimating the map quality for each node, which enables the robot to remove sufficiently informed nodes from the Quadtree. Two conditions are defined for the estimation of the map quality; entropy and maximum level of Quadtree. The proposed approaches were successfully implemented to our mobile robot equipped with sonar sensors for constructing a sonar map of an unknown environment in a real world.

Our contribution tackles the problem of learning to understand anaphoric references in the context of robotic machine learning; e.g. Get the large screw. Put it in the left hole. Our solution assumes the probabilistic theory of learning spelt out in earlier publications. Associations are formed probabilistically between constituents of the verbal command and constituents of a presupposed internal representation. The theory is extended, as a first step, to anaphora by learning how to distinguish between incorrect surface depth and the correct tree-structure depth of the anaphoric references.

A number of techniques, grouped under the umbrella term of usability inspection, have been devised in recent years to help software developers focus more closely on the needs of end-users during the development of interactive applications. Central to the use of such methods is the question of how proposed designs can be tested before a complete system has been implemented. This paper discusses the role of high and low-fidelity prototypes used for this purpose in software development and proposes the use of a “variable-fidelity” prototype as an aid to the development of interactive robotic systems. The approach is outlined and illustrated through the example application of an Interactive Robotic Visual Inspection System (IRVIS) developed for use by a person with limited physical manipulation skills.

This article deals with the kinematic calibration of the Delta robot. Two different calibration models are introduced: The first one takes into account deviations of all mechanical parts except the spherical joints, which are assumed to be perfect (“model 54”), the second model considers only deviations which affect the position of the end-effector, but not its orientation, assuming that the “spatial parallelogram” remains perfect (“model 24”). A measurement set-up is presented which allows to determine the end-effector's position and orientation with respect to the base. The measurement points are later be used to identify the parameters of the two calibration model resulting in an accuracy improvement of a factor of 12.3 for the position and a factor of 3.7 for the prediction of the orientation.

A person with a moderate or severe motor disability will often use specialised or adapted tools to assist their interaction with a general environment. Such tools can assist with the movement of a person's arms so as to facilitate manipulation, can provide postural supports, or interface to computers, wheelchairs or similar assistive technologies. Designing such devices with programmable stiffness and damping may offer a better means for the person to have effective control of their surroundings.

This paper addresses the possibility of designing some assistive technologies using impedance elements that can adapt to the user and the circumstances. Two impedance elements are proposed. The first, based on magnetic particle brakes, allows control of the damping coefficient in a passive element. The second, based on detuning the P-D controller in a servo-motor mechanism, allows control of both stiffness and damping. Such a mechanical impedance can be modulated to the conditions imposed by the task in hand. The limits of linear theory are explored and possible uses of programmable impedance elements are proposed.

This is an extension of previous work which used an artificial neural network with a back-propagation algorithm and a lookup table to find the inverse kinematics for a manipulator arm moving along pre-defined trajectories. The work now described shows that the performance of this technique can be improved if the back-propagation is made to be adaptive. Also, further improvement is obtained by using the whole workspace to train the neural network rather than just a pre-defined path. For the inverse kinematics of the whole workspace, a comparison has also been done between the adaptive back-propagation algorithm and radial basis function.

Assembly robots that use an active camera system for visual feedback can achieve greater flexibility, including the ability to operate in an uncertain and changing environment. Incorporating active vision into a robot control loop involves some inherent difficulties, including calibration, and the need for redefining the servoing goal as the camera configuration changes. In this paper, we propose a novel self-organizing neural network that learns a calibration-free spatial representation of 3D point targets in a manner that is invariant to changing camera configurations. This representation is used to develop a new framework for robot control with active vision. The salient feature of this framework is that it decouples active camera control from robot control. The feasibility of this approach is established with the help of computer simulations and experiments with the University of Illinois Active Vision System (UIAVS).